Re: DAB+/Wi-Fi radios due by end of the year

Boltar wrote:
On Aug 31, 9:06 pm, Richard Evans <R.P.Evans.NoS...@xxxxxxxxxxxx>
wrote:
MP3 performs a discrete cosine transformation, which could be done
using floating point, however it is also possible to avoid the use
of floating point by using a simple integer method. BTW. I say this
based on my experiences of creating jpeg decoders (jpeg also uses
DCT, and I assume that the techniques would be similar with mp3).

For those of us where cosine means the length of one side of a right
angled triangle depending on the angle , what exactly is a DCT and how
is it used?


It's used on MP3, AAC/AAC+, MPEG-2 video, MPEG-4 video, JPEG, and so on.

Its use in audio codecs consists of taking a vector of time-domain audio
samples then the DCT is taken (actually the modified DCT (MDCT)) of the
vector of samples to produce the frequency domain representation - the
transform is lossless, so it's just a different representation of the same
information.

To achieve compression codecs like AAC and MP3 simply encode the frequency
domain samples with fewer bits (or zero bits) than are needed for a lossless
representation - and they do that by taking advantage of frequency and time
masking.

You've heard of the FFT, right? The DCT is very similar to that apart from
the fact that the FFT of a signal produces complex (real + j*imaginary, I
don't mean complex as in "difficult") samples, whereas the DCT outputs
real-valued samples.



--
Steve - www.digitalradiotech.co.uk - Digital Radio News & Info


.

Relevant Pages

  • Re: C++ Calculate wrong answer.
    ... > I'm trying to calculate a very simple Discrete Cosine Transform (DCT), ... (snipped for brevity) ... Your main problem is that in C++, when you divide an integer by an integer, ... Use a floating point value or just do the divide ...
    (comp.lang.cpp)
  • Re: DAB+/Wi-Fi radios due by end of the year
    ... floating point, however it is also possible to avoid the use of floating ... experiences of creating jpeg decoders (jpeg also uses DCT, ...
    (alt.radio.digital)
  • Re: FFT VS DCT
    ... I see that all are asking why DCT is better than FFT or DFT. ... compared to high frequency energy concentration. ... tool to transform and to get maximum information out after quantisation. ...
    (comp.dsp)
  • Re: FFT VS DCT
    ... It's true if points in the original data have a correlation with their ... Where FFT is prepered than DCT? ... For image processing the DCT boundary conditions are less noticeable. ...
    (comp.dsp)
  • STUPIDENT::Re: FFT vs DCT
    ... Raeldor wrote: ... Is DCT basically the same as an FFT that operates on real numbers? ... 1d array of numbers? ...
    (comp.dsp)